I have been thinking a lot lately about the problem of expertise. By the problem of expertise, I mean how people who know better should relate to those who don't. Whether you are a physician or a physicist, this issue comes up a lot. People want the opinions of educated people -- pundits of various stripes proliferate -- but they do not always follow those opinions in their personal lives. Further, nearly every controversial scientific issue today involves some element of the knowers trying to impose their views on the know-nothings (or at least those who know significantly less). How should those experts approach others? What is the relevance of their special knowledge in the world?
The purpose of this post is to examine the meaning of expertise and its purpose in society. At the end, I also propose a solution to help experts deal with the lay public in a way that doesn't inspire acrimony.
For intelligent, educated people expertise is a problem. Many people reading this blog have suffered through an overwhelming amount of education. By virtue of this education, they have gained knowledge and skills that are tantamount to magic to the uninitiated. By virtue of their special skills, these experts are separated -- socially divided -- from the lay public. The problem of expertise is not that this special knowledge is undesirable. Special knowledge is both socially useful and personally satisfying. Rather the problem of expertise is a social problem, namely how those who know better should relate to those who don't.
People don't like to be governed, don't like to be told what to do, don't like a know-it-all. There is no need to belabor this subject. It is a cliche, but for experts it also a regular source of conflict with family, friends, and strangers. When a physician see someone engaging in unhealthy behavior, her impulse is to say: "You do realize what the consequences of this will be, don't you? You realize the self-destruction you're engaging in?" The physician is probably correct. She has seen the pathological slides, read the papers, and seen similar cases in clinic. On the other hand, the likely response in the poorly behaving person is annoyance if not rage: "How dare you treat me like a child? How dare you suggest that I am unhealthy?" The recipient of the criticism is also right. The physician is likely unfamiliar with the extenuating circumstances that resulted in that behavior. Furthermore, negative behaviors are not clinical destiny. Certain habits make disease more likely, but they do not usually guarantee that disease will occur.
How is this conflict between the two to be resolved? How should an experts relate to others who do not share their expertise? I argue that the problem of expertise is soluble, but that it requires dividing expertise into two separate sets of knowledge. Only one of them is true expertise -- although many assume that one implies the other.
The first type of expertise is expertise in the sense most people use the word. I will call this descriptive expertise. A descriptive expert has special knowledge about reality that is not generally distributed throughout the public. The physician understands the physiology of the body, the functions of different organ systems, and the molecules involved in facilitating life processes. The geologist understands the movements of rocks over long time scales; the physicist understands mechanics of moving bodies and the components of matter; the economist understands the behavior of prices as a consequences of changes in supply and demand. (That their expertise is limited by what we collectively know about these disciplines is irrelevant. A limited expert still knows a great deal more than the lay person.) The descriptive expert is an individual most qualified to answer inquiries into matters of fact.
The second type of expertise is expertise in the implications of facts. I will call this normative expertise. A normative expert understands facts as per descriptive expertise, but they also claim to understand the implications of those facts for people's lives. The normative expert judges human behavior or political policy (which is human behavior in aggregate) based on their perception of the logical consequences of those behaviors/policies vis-a-vis accepted facts. Their facts provide an ethical yardstick for guiding wise behavior. The normative expert is an individual who answers inquiries into matters of ethics and politics.
Before I raise too many eyebrows, let me argue two points with respect to normative expertise. First, this archetype is not irrelevant because an astonishingly large number of experts of the first type claim expertise of the second type -- at least with respect to their domain of understanding. Global warming scientists make claims not only on matters of fact -- global warming exists, expect temperature rises of X degrees, etc. -- but on matters of policy -- Kyoto is good, accomodation strategies are unwise, etc. Physicians make claims about facts -- trans-fats are bad for you -- and on what law should be -- trans-fats should be banned. Economists mix statements on how markets work -- increased demand will raise prices for oil -- with statements about equity -- it is unfair for oil companies to make so much money.
I am not suggesting that all experts do this or that some of them do it all the time. What I am suggesting is that the cross-over of descriptive expertise into normative expertise is quite prevalent. This is understandable. Descriptive experts -- including scientists -- are model-builders. They claim (whether this claim is warranted or not is another issue) predictive power over their model. They are also people, and people have principles and opinions. It is difficult to resist the temptation to mix one's opinions with one's perception of the facts.
The second point I would make is that -- in spite of the claims of some practitioners -- there is no reason for us to believe that normative expertise exists. Allow me to defend this proposition.
No matter how expert, no one has access to perfect information about what makes another person happy. Happiness is a non-transitive property of human life. Some may be perfectly happy working a minimum wage job their whole lives and shopping antique fairs on the weekend. Some will only be happy -- and maybe not even then -- living in luxury on 5th Ave. A child with terminal cancer can be happy after a day's play; a professional athlete who is the focus of public adulation may not be. Because happiness is not transitive -- what makes one person happy may not make another person happy -- all normative statements are necessarily circumscribed. When a normative expert claims that a particular behavior is wise, he predicates that claim on what he perceives as good -- usually an extension of what makes him happy.
You might respond: the expert could just ask. They could survey the individual to determine what makes her happy. Then the behavior that results in the greatest happiness could be established. Setting aside the issue of whether reliable information could be acquired in this manner -- people are not in my experience reliable witnesses as to what makes them happy -- I argue that getting this information is impossible.
Let's take the example of an individual. Say I am someone's physician. We are having an argument over whether or not they should smoke. Now I, as a physician, know for certain that smoking will increase their risk for a variety of very serious medical disorders. What I do not know are their reasons for smoking. It is very likely that they are physically addicted to nicotine. It is also very likely that there are underlying psychological factors such as responding to stress or the desire to fit into a social group. These encourage the person to make a poor health choice. It is unlikely that I will be able to enumerate much less understand all of these factors. I am not that person. And the relative weight of these factors figures into that person's decision whether or not to smoke. It figures into their decision how much to smoke. How much smoking is OK for that person? The calculation of what makes that person happy includes factors that are shielded from me as a physician. I cannot speak to these factors. What I can do is make clear in explicit terms what the likely consequences of their actions will be, but after that I can only leave them to weigh the consequences and the benefits as they see fit.
Don't think that is a good example? Consider a patient dying of cancer. They are likely in a great deal of pain. I can detail to them their treatment options. I can compare those options with means of limiting their pain. But I cannot truly understand their pain, so I cannot make the decision of whether to fight for them. It is the nature of subjective experience that we cannot make normative decisions for others. I might add that this is the core principle behind the idea of patient autonomy: the idea that a physician can help inform the patient but cannot make their medical decisions for them. It also illustrates a problem with normative expertise. You cannot claim normative expertise even when dealing with a single individual because you don't know their preferences precisely.
The problem of information about preferences becomes even more dramatic -- and fatal to normative expertise -- when we consider individuals in aggregate. Consider the example of the supermarket. Everyday (or every week) people go to the supermarket to purchase the food that they would like to eat. Could this system of supermarkets be ably supplanted by a normative expert -- a sort of benevolent dictator of food distribution? To compare the relative success of these two system, let's define success as individuals getting the food that they want to eat each day. To establish what people want, the dictator sends surveys to individuals about what food deliveries they would like on each given day. The dictator arranges for the production of these foods by purchasing the appropriate amounts from the particular suppliers. What would happen to such a system in the real world? First many individuals would change their minds about what they wanted during the course of the day. That would render the survey ineffective at establishing their preferences. Even if surveying was performed more regularly, you get into a problem of food production. How many potatoes should we make? How much do people like potatoes? Potatoes take time to produce, and rapid changes in individual preference cannot result in instantaneous changes in food delivery. Furthermore, there is a problem of payment. Let's say that under the benevolent dictator scenario, the individual turns over whatever money they were going to spend on food to the expert in return for food stuffs. The problem is that food preferences are dissociated from price. Say more people want chicken over a particular month. The supermarket changes the prices to accommodate for increased demand, bringing the quantity demanded in line with the quantity supplied and preventing shortages. A benevolent dictator cannot respond to changes in demand with changes in price, and even if he could it wouldn't be as quick as in the price system. This is the information function of price system. Prices carry information as to people's preferences for goods and attempts to collect this information through means other than prices leads to collosal waste and inefficiency. No person or group of people can accomplish this information distribution function as quickly and effectively as prices do. (For more information on the information function of prices, read I, Pencil by Leonard Read.)
Now, you may think that my examples of the limits of information are irrelevant to problems of normative expertise in our society. No one is advocating (anymore) that food be distributed by a benevolent dictator. However, consider global warming. Many have argued as a global warming abatement strategy that we impose emissions caps. The problem with these caps is that they divorce the distribution of the license to emit from the relative preferences of the individuals doing the emitting. All individuals want to use energy, but you have no idea how much they want to emit. Maybe it is more desirable to one person whereas another would agree to go without. Any entity distributing such licenses to emit would lack the information necessary to make that decision -- information about relative preferences. The benefit of a price system is that it establishes relative preference via who is willing to pay the most. Those who are willing to pay the most can emit the most. (As an aside, this is why I think a carbon tax is a far more efficient way to regulate carbon emissions than a cap system.) Now there are some ways to put the price system back into the cap policy. You could have a cap-and-trade market or a public auction. This would reintroduce the concept of relative preference, but note that it also limits the powers of the normative expert. The expert no longer can dictate the relative value of carbon emissions; they can no longer dictate who should and should not emit.
In the above paragraphs, I argue that the normative "expert" lacks the necessary information with regards to preferences to make informed judgments. This is the first problem with normative expertise, but there are many others. The Right regularly argues against normative expertise by arguing against descriptive expertise. They doubt the ability of the expert in understanding the facts and producing models with predictive capacity. (I discussed this in detail here.)
Another problem with the normative expert is that they are necessarily use their own ethical principles in measuring the relative value of two scenarios. People differ with respect to wants and needs. I argued above that a benevolent dictator lacks the necessary information for informed judgment, but this assumes that the dictator is benevolent, i.e. makes the decision that most people want. On the contrary, the normative expert might be determined to impose their minority principles on others. In this case, the normative expert is predicating their expertise on the inherent rightness of their values. Establishing the rightness or wrongness of facts is pretty easy; establishing the rightness or wrongness of different principles is much more difficult. You get into the same information problem as above. What might be a good principle for you may not be for me.
The purpose of this argument is to illustrate that while descriptive expertise certainly exists, normative expertise is much more questionable.
To extend this line of reasoning, I also argue that the problem that most members of the public have with science is their perception that scientists often pretend to normative expertise. Not only do people perceive an element of bullying in this pretense, they intuitively grasp the shaky ground on which the assumption of such expertise rests. Scientists and other experts would receive a much better reception if they stuck with descriptive expertise rather than indulging in normative claims.
Let's set aside arguments of the degree to which particular scientists indulge in this behavior. None of us is entirely innocent of this pernicious vice, and it is a pernicious one. Theoretical overreach into areas not within your knowledge is one thing. It can be entirely solipsistic; no one else need be involved. That is just scientism. Claiming normative expertise is not only pernicious because it rests on an unfounded claim, it judges others by an unfair standard. It also has an annoying habit of supporting arguments about how other people's rights should be abridged. "We know better" often morphs into "others should be stopped."
As a practical matter, let's focus on how to divorce descriptive from normative expertise. This is admittedly a fine line to draw for oneself, and it takes effort and practice to walk it. I am hardly innocent of injecting my opinions into statements of fact on these pages.
How do this work in practical terms? I am in a sense an expert -- though this is a relative statement because I still have a great deal of education left to endure. In my interactions with family members, they often inquire about my medical opinion, unschooled though it may be. However, in providing them answers to their questions I try to speak only to matters of fact -- this behavior, this illness or this intervention is likely to produce the following results -- rather than matters of personal principle. I would do well to remember my personal state. Anyone who knows me well knows that my personal life is a matter of continuing travesty, an ongoing road-show of farcically poor judgment. I am barely qualified to take care of myself, much less inflict my value judgment on others.
I urge scientists and other experts to adopt a policy of divorcing normative and descriptive expertise. As aspiring experts it is wise for us to use considerable caution in implying that our expertise vis-a-vis facts implies a better judgment with respect to values. Considerable acrimony between experts and the public would be avoided if the experts more carefully distinguished between descriptive and normative expertise, and knew well which one they possess.
Well, that's a very nice analysis, but there is a much simpler route to the endpoint: Everyone should just shut the fuck up and do what I fucking tell them!!
Seriously, that is a very nice post, which you clearly worked hard on.
Very nice post.
As a therapist people ask me all the time what they should do. In my style of therapy this is avoided -- which leads to all kinds of frustration. But I think part of what's nice about this type of therapy is that it undervalues what you're calling 'normative expertise'.
I'm willing to examine patients motivations and decisions and what they mean to them, but I'm unwilling to say that I know better than them when it comes to living their own lives. In the end, people come to appreciate that I don't tell them what to do.
"We know better" often morphs into "others should be stopped."
Nice one Jake. Although you choose medicine for your examples the principles generalize well.
Well enough for individual case. But at the end of the day political decisions have to be made, and we resist "normative experts" diagnoses for good and bad reasons. The good is that many "normative experts" are not so expert at all - just see the field of economics for that.
But the bad is exactly this confusion between individual judgement and group judgment. When we make a judgment as a group, we should of course rely primarily on those who understand the implications of the current state of affairs. To pretend that we are all equally capable of understanding the political and moral ramifications of political judgments in every field is the height of anti-intellectualism, and ultimately self-destructive.
It is silly to tell others what to do in their own lives as a personal policy; but it is insane to not delegate the decision of social restraints. I suggest that the problem is exactly the reverse of what you posit: it's not that experts need to divorce themselves from advancing political positions - they already do so. No, as a community we need better judgment in who is a normative expert; but we need scientists and other experts to take a greater political role.
Otherwise, we end up with our current crop of demagogues in power, who revel in their lack of expertise.
I seem to recall that you were taking an introductory economics course. Well, the argument you just made, while not without merit, is an example of a micro/macro fallacy. People may resent or ignore the personal advice (micro) of an expert, but it doesn't follow that such personal resentment means that experts should be ignored in the public (macro) realm.
This is just standard libertarianism. Nothing wrong with that, but don't try to dress it up as something else.
As a political philosopher (who wandered over from reading Evolving Thoughts), I have to disagree in two respects. First, normative judgements are essential for your post; you can't do without them, and so you shouldn't do without them. And second, your argument that normative judgements are impossible is logically invalid. And let me apologise in advance for the length of this comment. :-)
Sentences like `None of us is entirely innocent of this pernicious vice, and it is a pernicious one' and `Claiming normative expertise ... judges others by an unfair standard' essentially involve normative judgements. Furthermore, the primary conclusion you're trying to establish here is that descriptive experts should avoid the temptation to play normative experts. That's a normative judgement. If you think that we should not be making normative judgements, it follows immediately that you shouldn't be making the judgement that is the entire point of this post. You've completely failed to speak only to matters of fact.
You might claim that those sentences were just mistakes -- you gave into temptation and made a normative judgement. How can you repair the post? Well, you might strike through those sentences, or tell your readers to ignore them. But then the post is pointless. The reason you wrote the post was to argue for the normative judgement that we shouldn't be making normative judgements. Without that conclusion, you have no reason to write the post, and no-one else has any reason to read it.
Alternatively, you might replace those sentences with purely descriptive sentences. For example, your conclusion might just be the claim that making normative judgements are impossible on epistemological grounds. (Or something slightly weaker but significantly more complicated.) But, again, then the post is pointless on its own. The fact that the performance of some action A is impossible is interesting, significant, worth considering, etc., only to someone who would like to do A. You wouldn't write a post in which you reported the number of grains of sand on a beach unless you had some reason to care about how many grains of sand were on that beach. And reasons, ends, purposes, and desires have normative content.
So unless you reasonably assumed that your audience wants to do certain things (namely, make normative judgements), there was no reason for you to write this post. But then your argument falls apart, because you were trying to show that we can't reasonably assume that the people around us want to do certain things.
Which brings me to my second critique, which is of your argument itself. I'll readily grant that we can't know with certainty what other people's interests, desires, and plans are, for exactly the reasons you give here. But that's neither necessary nor sufficient to make fallible and provisional normative judgements. For those, all we need are some fallible and provisional descriptive generalisations about what people's interests are: Since people generally don't like dying at a relatively young age in an extremely painful way, people shouldn't smoke. Since people generally don't like laying in bed awake all night with their minds racing, people shouldn't drink coffee later than six in the evening. If we do happen to meet someone who is perfectly fine with dying at a relatively young age (say, because they're already dying of incurable leukemia), then our normative judgement has been falsified, and we will want to revise our judgement to include, say, a ceteris paribus clause. But this just means that your argument is logically invalid -- the premisses are true, but the conclusion, strictly speaking, is false.
Furthermore, these descriptive generalisations are exactly the ones we already use in order to get around in the world. I can't individually trust the thousands of strangers with whom I share the highway. I have to assume that they generally don't want to die, reason to the intermediate conclusion that they're not going to deliberately cause an accident, and then conclude that, for example, an oncoming car won't suddenly swerve into my lane. So we can make fallible and provisional normative judgements -- just by virtue of successfully navigating the human part of the world, we have all the resources we need for normative judgements.
It's true that some of these assumptions are bad ones, and it's wrong to assume that everyone values exactly the same things as I do to exactly the same degree. But that doesn't mean that normative judgements are impossible. As John Dewey argued 80 years ago, it just means that normative judgements, like descriptive judgements, have to subjected to test and revision.
I've had enough education on negotiation to be dangerous. In the broad sense all human interactions are negotiations. Politics is negotiations, etc. One would think outcome of negotiation would rest on expert advice. This in not the case. Truth, facts, expert advice, moral high ground, etc. are only negotiating tools. How well these tools are employed by each side is one of a number of factors which determine the outcome of the negotiation. It is very common for environmental negotiations to come to conclusions which appear to ignore the facts of the matter.
I see scientific expertise as saying how the universe is. However, it is not the job of the scientist to say what is good or bad. However, the expert should be best qualified to estimate what the results of various possible actions could be. So, in that sense, I think normative advice is appropriate.
your argumentation would be fine, but IMO your assumptions about John's claims are wrong.
False: Normative judgements are bad.
Right: Claiming normative *expertise* is bad.
What does that mean ? It means that people may freely use normative judgements; John's post is not contradictory. But John said that there are *no* privileged persons which, when asked by a person or a group, are able to give consistently more *pleasing* answers or point out a better, *more happiful* way than other persons.
Example: The use of a gun. I may give the normative judgement that shooting a person is always a very unpleasant experience and many people will agree. This normative judgement is, if I understand Johns position correctly, ok.
But if you ask: If I surprise a burglar, would I feel better to have a gun or not ? then John says that my or your answer is exactly as good as the advise of a psychiatrist or a self-defense instructor.
If you claim: "You should not use a gun because the usage will cripple you psychically !" then you are claiming normative expertise.
The problem is: You don't know what will happen.
If you don't have a gun:
Perhaps the person will talk out the burglar to not use violence or escapes and say: "Man, that was a good decision !". But equally possible is that the person loses all self-confidence or is traumatized because he couldn't defend himself against abuse and says: "Never, ever again !".
The same with a gun: Either you defend your property or your family or you handle the situation so good by arresting the thief that you say: "Wow, I am so glad that I had my gun". But equally possible is that you lose your nerve, shoot the burglar down and realize that you are severely traumatized by the dying man.
Wrong: Normative judgements are impossible.
Right: The informations which are necessary to guide a person to archieve the best personal solution are impossible to get.
Again the gun: You can get probabilistic answers what happens to people who (don't) use a gun. You can analyze
the background (peaceful/combat-ready -- social worker/soldier) of the person. So Johns claim does not deny that you may get a clue about the outcome of your decision.
But still there are many parameters which you can't get to predict successfully the outcome of an altercation (and even if you could, the problem is that the person may not believe in retrospect that it was really the best outcome).
Sorry, Jake. John Wilkins linked to it and I screwed up.
Okay, I think I need to get a better understanding of what claiming expertise amounts to. Based on your comment and the post, here's one definition:
(1) P has expertise about X matters (or `X expertise') if and only if P's knowledge claims concerning X matters are infallible. (Less jargony, if and only if everything that P knows concerning X is guaranteed to be true.)
Plug in `descriptive' for X to get a definition of descriptive expertise -- that's if and only if P's knowledge claims concerning descriptive matters are infallible. When we actually talk about expertise, we narrow things a bit -- say, to (descriptive) neuroscience expertise, or (descriptive) particle physics expertise, and so on. P has expertise about neuroscience matters (respectively, particle physics matters) if and only if P's knowledge claims concerning neuroscience (particle physics) matters are infallible. And similarly with `normative'. P has normative expertise if and only if P's knowledge claims concerning normative matters are infallible.
If this is what you and/or our host mean by expertise, then I'll concede your arguments. It certainly is the case that no-one's knowledge claims concerning normative matters are infallible, and so no-one has normative expertise, QED. But, for reasons that go back to Hume, no-one has descriptive expertise (of any kind) either. Our scientific knowledge claims aren't infallible, and I don't know of anyone who takes the history or philosophy of science of the last 150 years seriously who thinks otherwise.
So if that's what you (meaning, again, you or our host) mean by expertise, the arguments and the claims are fine as far as they go. But then I don't think the claims amount to anything we haven't recognised for a long time already.
Here's a weaker definition of expertise:
(2) P has expertise about X matters if and only if P's knowledge claims concerning X matters are reliable. (`Reliable' is a technical term in contemporary epistemology. Very loosely, this definition amounts to `P's knowledge claims concerning X are pretty good, but aren't guaranteed to be true'.)
This is closer to what I had in mind when I was writing my last comment, and it does seem true to say that scientists have expertise in their respective areas. But then you can't argue that, since normative knowledge claims are fallible, normative expertise is impossible.
Generally speaking, you want, I think, a definition of expertise such that you can show (a) some people have descriptive expertise and (b) no-one has normative expertise. You want to put these two kinds of expertise on two different levels. In both of the definitions I've given, descriptive and normative expertise come out on the same level. And I can't think of another definition of expertise that doesn't put them on the same level (at least, without begging the question). Can you?
Very well-written. And Noumena and TSK help to clarify things even more.
Random thought though: I think that this relates to the "enabler" foolishness, because the prescriptions on normative expertise go both ways - Mooney and Nisbet would do better to avoid trying to invoke their normative views on scientists, that is. As a result, they're alienating just about everyone on the science side of discussion over the movie Expelled, just as you suggest that scientists can alienate non-experts.
Noumena, I believe you are right as to the more radical form of the argument made by Young. People with descriptive expertise in a field usually have better choices relating to that field.
But this, I think, misses the point.
What I do find correct about Young's post, is a weaker version of the claim. I don't think anyone (other that a zealot libertarian) would claim that Knowledge is irrelevant to Normative judgments. But it is good practice to disentangle the two.
I think about my workplace: my boss needs to understand what I am doing. He has the Normative right to decide how I do it. What he lacks is knowledge. After all, I am doing the actual work, I know the small print because I do it. So I usually know better what should be done. But he has the last word. Now, my boss thinks I should do A, and I thing B is better. I have my arguments. As long as I keep my arguments in order, and contrive to separate them from the judgments I make, my boss is far more likely to listen. When I entangle it all, and just "make the claim from expertise", it usually ends with the boss telling me what to do, and all my expertise can go down the drain.
Yep, I can, because your definition of expertise is missing an important point: It is *not* expertise if P's knowledge claims concerning X is reliable, it is if P's predictions are by a significant margin *better* than Q's non-expert predictions.
If the task is identifying color, P can get a success rate of 99.7% and would be an expert by your definition because his predictions are very reliable, but I as non-expert can get 99.5%, too. People would agree that despite his reliablility P does not deserve the status "expert".
If on the other side a fault in a complicated network must be found and a random choose would find it in 1:1 000 000 cases, an educated non-expert can get it down to 1:1000 and expert P down to 1:100, then P's predictions are still extremely unreliable, but by a magnitude better than non-experts so his status as an expert is legitimate.
You don't need to argue about the absolute truth value or reliability and therefore your meta-argument of beating
Jakes claim with its own preconditions breaks down. To attack the conclusion you must show that Jakes claim of
an non-existing normative expertise is not holdable; that people may build up a much more precise estimation of normative judgements than other people. But Jake points out that we cannot acquire the necessary information because people are unsure about their inherent values.